Market Prediction

Investment firms, hedge funds and even individuals have been using financial models to better understand market behavior and make profitable investments and trades. A wealth of information is available in the form of historical stock prices and company performance data, suitable for machine learning algorithms to process.

For this project, your task our task is to build a stock price predictor that takes daily trading data over a certain date range as input, and outputs projected estimates for given query dates.

Table of Contents

• Retrive Data
• Build an LSTM model
• Build a GRU model
• References

# import libraries
import pandas as pd
from pandas_datareader import data as pdr
import numpy as np


import plotly.express as px
import plotly.graph_objects as go
import matplotlib.pyplot as plt
import seaborn as sns

from sklearn.metrics import mean_squared_error
from sklearn.model_selection import GridSearchCV
from sklearn.preprocessing import MinMaxScaler

import datetime
import time
import math


import torch
import torch.nn as nn

Data Preparation

start_date = datetime.datetime(2015, 1, 1) 
end_date   = datetime.datetime(2020,12,31)


#Tickers : GM AMZN BTC-USD
df = pdr.get_data_yahoo('GM', start=start_date, end=end_date).reset_index()

price = df[['Adj Close']]
scaler = MinMaxScaler(feature_range=(-1, 1))
price['Adj Close'] = scaler.fit_transform(price['Adj Close'].values.reshape(-1,1))

/opt/conda/lib/python3.6/site-packages/ipykernel_launcher.py:10: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

def split_data(stock, lookback):
    """
    Split the daily data in a training set an a test set

    Parameters
    ----------
    stock : DataFrame
        Daily data 
    lookback : int
        Lookback window     

    Returns
    -------
    x_train, y_train, x_test, y_test : arrays
        The training sets and testing sets
    """
    data_raw = stock.to_numpy() # convert to numpy array
    data = []
    
    # create all possible sequences of length seq_len
    for index in range(len(data_raw) - lookback): 
        data.append(data_raw[index: index + lookback])
    
    data = np.array(data);
    test_set_size = int(np.round(0.2*data.shape[0]));
    train_set_size = data.shape[0] - (test_set_size);
    
    x_train = data[:train_set_size,:-1,:]
    y_train = data[:train_set_size,-1,:]
    
    x_test = data[train_set_size:,:-1]
    y_test = data[train_set_size:,-1,:]
    
    return [x_train, y_train, x_test, y_test]

lookback = 10 # choose sequence length
x_train, y_train, x_test, y_test = split_data(price, lookback)


x_train = torch.from_numpy(x_train).type(torch.Tensor)
x_test = torch.from_numpy(x_test).type(torch.Tensor)
y_train_lstm = torch.from_numpy(y_train).type(torch.Tensor)
y_test_lstm = torch.from_numpy(y_test).type(torch.Tensor)
y_train_gru = torch.from_numpy(y_train).type(torch.Tensor)
y_test_gru = torch.from_numpy(y_test).type(torch.Tensor)

if torch.cuda.is_available():
    device = torch.device("cuda")
else:
    device = torch.device("cpu")
    
print(device)

#Models parameters
input_dim = 1
hidden_dim = 64 #32 64 128
num_layers = 2 #2 , 3
output_dim = 1
num_epochs = 500 #100 250 500 1000

cuda

Build an LSTM model

class LSTM(nn.Module):
    def __init__(self, input_dim, hidden_dim, num_layers, output_dim):
        super(LSTM, self).__init__()
        self.hidden_dim = hidden_dim
        self.num_layers = num_layers
        
        self.lstm = nn.LSTM(input_dim, hidden_dim, num_layers, batch_first=True)
        self.fc = nn.Linear(hidden_dim, output_dim)    
        
    def forward(self, x):
        h0 = torch.zeros(self.num_layers, x.size(0), self.hidden_dim).requires_grad_().to(device)
        c0 = torch.zeros(self.num_layers, x.size(0), self.hidden_dim).requires_grad_().to(device)
        out, (hn, cn) = self.lstm(x, (h0.detach(), c0.detach()))
        out = self.fc(out[:, -1, :]).to(device)
        return out

model = LSTM(input_dim=input_dim, hidden_dim=hidden_dim, output_dim=output_dim, num_layers=num_layers)
criterion = torch.nn.MSELoss(reduction='mean')

#optimiser = torch.optim.SGD(model.parameters(), lr=0.01)
optimiser = torch.optim.Adam(model.parameters(), lr=0.01)

model.to(device)
criterion.to(device)
x_train = x_train.to(device)
y_train_lstm = y_train_lstm.to(device)

hist = np.zeros(num_epochs)
start_time = time.time()
lstm = []

for t in range(num_epochs):
    y_train_pred = model(x_train).to(device)    
    
    loss = criterion(y_train_pred, y_train_lstm)
    print("Epoch ", t, "MSE: ", loss.item())
    hist[t] = loss.item()    
    
    optimiser.zero_grad()
    loss.backward()
    optimiser.step()
    
training_time = time.time()-start_time
print("Training time: {}".format(training_time))

Epoch  0 MSE:  0.10327287763357162
Epoch  1 MSE:  0.10609275102615356
Epoch  2 MSE:  0.084849514067173
Epoch  3 MSE:  0.07953985780477524
Epoch  4 MSE:  0.06514491885900497
Epoch  5 MSE:  0.029029162600636482
Epoch  6 MSE:  0.018505150452256203
Epoch  7 MSE:  0.12133926153182983
Epoch  8 MSE:  0.022022387012839317
Epoch  9 MSE:  0.034502267837524414
Epoch  10 MSE:  0.04111925885081291
Epoch  11 MSE:  0.045831792056560516
Epoch  12 MSE:  0.049516815692186356
Epoch  13 MSE:  0.051705021411180496
Epoch  14 MSE:  0.0520317442715168
Epoch  15 MSE:  0.05007021874189377
Epoch  16 MSE:  0.04515652358531952
Epoch  17 MSE:  0.036432500928640366
Epoch  18 MSE:  0.023690905421972275
Epoch  19 MSE:  0.011560875922441483
Epoch  20 MSE:  0.01673886924982071
Epoch  21 MSE:  0.0173213891685009
Epoch  22 MSE:  0.009943884797394276
Epoch  23 MSE:  0.017121853306889534
Epoch  24 MSE:  0.017338145524263382
Epoch  25 MSE:  0.009654054418206215
Epoch  26 MSE:  0.0052611446008086205
Epoch  27 MSE:  0.006763740908354521
Epoch  28 MSE:  0.010432498529553413
Epoch  29 MSE:  0.012153762392699718
Epoch  30 MSE:  0.010830311104655266
Epoch  31 MSE:  0.008078470826148987
Epoch  32 MSE:  0.005913082044571638
Epoch  33 MSE:  0.005378433037549257
Epoch  34 MSE:  0.0062453122809529305
Epoch  35 MSE:  0.0071733491495251656
Epoch  36 MSE:  0.006912760902196169
Epoch  37 MSE:  0.005903179757297039
Epoch  38 MSE:  0.0054878066293895245
Epoch  39 MSE:  0.00565411988645792
Epoch  40 MSE:  0.005462123546749353
Epoch  41 MSE:  0.004951308947056532
Epoch  42 MSE:  0.004792962688952684
Epoch  43 MSE:  0.005101779010146856
Epoch  44 MSE:  0.0054450989700853825
Epoch  45 MSE:  0.005403035320341587
Epoch  46 MSE:  0.0049084387719631195
Epoch  47 MSE:  0.004292689263820648
Epoch  48 MSE:  0.004057419020682573
Epoch  49 MSE:  0.004382362123578787
Epoch  50 MSE:  0.004800620023161173
Epoch  51 MSE:  0.0047332593239843845
Epoch  52 MSE:  0.00429498078301549
Epoch  53 MSE:  0.004010020289570093
Epoch  54 MSE:  0.00405229302123189
Epoch  55 MSE:  0.0042020706459879875
Epoch  56 MSE:  0.00423399405553937
Epoch  57 MSE:  0.004127885214984417
Epoch  58 MSE:  0.004010280594229698
Epoch  59 MSE:  0.0039786044508218765
Epoch  60 MSE:  0.003988081123679876
Epoch  61 MSE:  0.003933039028197527
Epoch  62 MSE:  0.003823513863608241
Epoch  63 MSE:  0.003776049939915538
Epoch  64 MSE:  0.003824274055659771
Epoch  65 MSE:  0.003863774472847581
Epoch  66 MSE:  0.003804313950240612
Epoch  67 MSE:  0.0036872613709419966
Epoch  68 MSE:  0.0036205435171723366
Epoch  69 MSE:  0.003640422597527504
Epoch  70 MSE:  0.0036781001836061478
Epoch  71 MSE:  0.003658491186797619
Epoch  72 MSE:  0.0035898007918149233
Epoch  73 MSE:  0.0035334117710590363
Epoch  74 MSE:  0.003518172074109316
Epoch  75 MSE:  0.003517123172059655
Epoch  76 MSE:  0.0034970585256814957
Epoch  77 MSE:  0.0034619851503521204
Epoch  78 MSE:  0.0034362394362688065
Epoch  79 MSE:  0.0034217468928545713
Epoch  80 MSE:  0.0033987327478826046
Epoch  81 MSE:  0.0033639115281403065
Epoch  82 MSE:  0.003336795140057802
Epoch  83 MSE:  0.0033262858632951975
Epoch  84 MSE:  0.0033164448104798794
Epoch  85 MSE:  0.003290680004283786
Epoch  86 MSE:  0.0032549300231039524
Epoch  87 MSE:  0.0032281929161399603
Epoch  88 MSE:  0.0032156326342374086
Epoch  89 MSE:  0.003202912863343954
Epoch  90 MSE:  0.0031789366621524096
Epoch  91 MSE:  0.003150469856336713
Epoch  92 MSE:  0.0031281677074730396
Epoch  93 MSE:  0.003110940335318446
Epoch  94 MSE:  0.0030915706884115934
Epoch  95 MSE:  0.003068988909944892
Epoch  96 MSE:  0.0030475202947854996
Epoch  97 MSE:  0.0030280023347586393
Epoch  98 MSE:  0.003006906481459737
Epoch  99 MSE:  0.0029837179463356733
Epoch  100 MSE:  0.002962325233966112
Epoch  101 MSE:  0.00294388341717422
Epoch  102 MSE:  0.002924472326412797
Epoch  103 MSE:  0.002901745028793812
Epoch  104 MSE:  0.0028787213377654552
Epoch  105 MSE:  0.00285856775008142
Epoch  106 MSE:  0.0028396756388247013
Epoch  107 MSE:  0.002818876411765814
Epoch  108 MSE:  0.002796490676701069
Epoch  109 MSE:  0.002774963853880763
Epoch  110 MSE:  0.0027546226046979427
Epoch  111 MSE:  0.0027338836807757616
Epoch  112 MSE:  0.002712367568165064
Epoch  113 MSE:  0.002691052621230483
Epoch  114 MSE:  0.0026700596790760756
Epoch  115 MSE:  0.002648686757311225
Epoch  116 MSE:  0.0026271298993378878
Epoch  117 MSE:  0.0026060682721436024
Epoch  118 MSE:  0.0025851137470453978
Epoch  119 MSE:  0.0025634451303631067
Epoch  120 MSE:  0.0025413997936993837
Epoch  121 MSE:  0.002519859466701746
Epoch  122 MSE:  0.0024986269418150187
Epoch  123 MSE:  0.0024769043084234
Epoch  124 MSE:  0.00245482474565506
Epoch  125 MSE:  0.0024330804590135813
Epoch  126 MSE:  0.0024115964770317078
Epoch  127 MSE:  0.00238981656730175
Epoch  128 MSE:  0.0023677637800574303
Epoch  129 MSE:  0.002345810877159238
Epoch  130 MSE:  0.0023239431902766228
Epoch  131 MSE:  0.002301996573805809
Epoch  132 MSE:  0.0022800967562943697
Epoch  133 MSE:  0.0022583564277738333
Epoch  134 MSE:  0.0022366882767528296
Epoch  135 MSE:  0.00221512233838439
Epoch  136 MSE:  0.002193776424974203
Epoch  137 MSE:  0.0021725890692323446
Epoch  138 MSE:  0.0021515486296266317
Epoch  139 MSE:  0.0021308877039700747
Epoch  140 MSE:  0.0021107173524796963
Epoch  141 MSE:  0.0020909467712044716
Epoch  142 MSE:  0.0020716742146760225
Epoch  143 MSE:  0.0020530743058770895
Epoch  144 MSE:  0.0020350839477032423
Epoch  145 MSE:  0.0020177073311060667
Epoch  146 MSE:  0.0020011684391647577
Epoch  147 MSE:  0.0019855087157338858
Epoch  148 MSE:  0.001970660872757435
Epoch  149 MSE:  0.001956736436113715
Epoch  150 MSE:  0.001943748677149415
Epoch  151 MSE:  0.0019316324032843113
Epoch  152 MSE:  0.001920461654663086
Epoch  153 MSE:  0.001910197315737605
Epoch  154 MSE:  0.0019007594091817737
Epoch  155 MSE:  0.0018921061418950558
Epoch  156 MSE:  0.0018841071287170053
Epoch  157 MSE:  0.0018767131259664893
Epoch  158 MSE:  0.001869786879979074
Epoch  159 MSE:  0.0018631998682394624
Epoch  160 MSE:  0.001856886432506144
Epoch  161 MSE:  0.0018506691558286548
Epoch  162 MSE:  0.00184451334644109
Epoch  163 MSE:  0.0018383247079327703
Epoch  164 MSE:  0.0018320315284654498
Epoch  165 MSE:  0.001825616112910211
Epoch  166 MSE:  0.0018190358532592654
Epoch  167 MSE:  0.0018123004119843245
Epoch  168 MSE:  0.0018053972162306309
Epoch  169 MSE:  0.001798370503820479
Epoch  170 MSE:  0.0017912328476086259
Epoch  171 MSE:  0.0017840288346633315
Epoch  172 MSE:  0.0017768022371456027
Epoch  173 MSE:  0.0017695720307528973
Epoch  174 MSE:  0.0017623926978558302
Epoch  175 MSE:  0.0017552823992446065
Epoch  176 MSE:  0.001748264068737626
Epoch  177 MSE:  0.0017413527239114046
Epoch  178 MSE:  0.0017345469677820802
Epoch  179 MSE:  0.0017278512241318822
Epoch  180 MSE:  0.0017212586244568229
Epoch  181 MSE:  0.001714757177978754
Epoch  182 MSE:  0.0017083346610888839
Epoch  183 MSE:  0.0017019727965816855
Epoch  184 MSE:  0.001695660874247551
Epoch  185 MSE:  0.0016893836436793208
Epoch  186 MSE:  0.0016831276006996632
Epoch  187 MSE:  0.0016768834320828319
Epoch  188 MSE:  0.0016706380993127823
Epoch  189 MSE:  0.001664385781623423
Epoch  190 MSE:  0.0016581222880631685
Epoch  191 MSE:  0.0016518420306965709
Epoch  192 MSE:  0.0016455440782010555
Epoch  193 MSE:  0.0016392271500080824
Epoch  194 MSE:  0.0016328906640410423
Epoch  195 MSE:  0.0016265381127595901
Epoch  196 MSE:  0.0016201696125790477
Epoch  197 MSE:  0.0016137883067131042
Epoch  198 MSE:  0.0016073979204520583
Epoch  199 MSE:  0.0016009998507797718
Epoch  200 MSE:  0.0015945985214784741
Epoch  201 MSE:  0.001588196842931211
Epoch  202 MSE:  0.0015817983075976372
Epoch  203 MSE:  0.0015754059422761202
Epoch  204 MSE:  0.001569023123010993
Epoch  205 MSE:  0.0015626528766006231
Epoch  206 MSE:  0.0015562985790893435
Epoch  207 MSE:  0.0015499631408602
Epoch  208 MSE:  0.0015436491230502725
Epoch  209 MSE:  0.0015373599017038941
Epoch  210 MSE:  0.0015310978051275015
Epoch  211 MSE:  0.001524865860119462
Epoch  212 MSE:  0.001518666627816856
Epoch  213 MSE:  0.0015125020872801542
Epoch  214 MSE:  0.0015063751488924026
Epoch  215 MSE:  0.0015002868603914976
Epoch  216 MSE:  0.0014942394336685538
Epoch  217 MSE:  0.0014882339164614677
Epoch  218 MSE:  0.0014822717057541013
Epoch  219 MSE:  0.0014763528015464544
Epoch  220 MSE:  0.0014704770874232054
Epoch  221 MSE:  0.0014646442141383886
Epoch  222 MSE:  0.001458853716030717
Epoch  223 MSE:  0.0014531039632856846
Epoch  224 MSE:  0.0014473930932581425
Epoch  225 MSE:  0.0014417197089642286
Epoch  226 MSE:  0.0014360809000208974
Epoch  227 MSE:  0.0014304752694442868
Epoch  228 MSE:  0.0014248999068513513
Epoch  229 MSE:  0.0014193531824275851
Epoch  230 MSE:  0.00141383265145123
Epoch  231 MSE:  0.0014083373825997114
Epoch  232 MSE:  0.0014028659788891673
Epoch  233 MSE:  0.0013974178582429886
Epoch  234 MSE:  0.0013919930206611753
Epoch  235 MSE:  0.0013865914661437273
Epoch  236 MSE:  0.001381214358843863
Epoch  237 MSE:  0.001375863328576088
Epoch  238 MSE:  0.0013705401215702295
Epoch  239 MSE:  0.0013652476482093334
Epoch  240 MSE:  0.0013599882367998362
Epoch  241 MSE:  0.0013547652633860707
Epoch  242 MSE:  0.0013495823368430138
Epoch  243 MSE:  0.0013444427167996764
Epoch  244 MSE:  0.0013393503613770008
Epoch  245 MSE:  0.0013343093451112509
Epoch  246 MSE:  0.0013293232768774033
Epoch  247 MSE:  0.0013243958819657564
Epoch  248 MSE:  0.0013195313513278961
Epoch  249 MSE:  0.0013147329445928335
Epoch  250 MSE:  0.0013100046198815107
Epoch  251 MSE:  0.0013053498696535826
Epoch  252 MSE:  0.0013007718371227384
Epoch  253 MSE:  0.001296274014748633
Epoch  254 MSE:  0.0012918595457449555
Epoch  255 MSE:  0.001287531922571361
Epoch  256 MSE:  0.0012832944048568606
Epoch  257 MSE:  0.0012791494373232126
Epoch  258 MSE:  0.0012751005124300718
Epoch  259 MSE:  0.001271150540560484
Epoch  260 MSE:  0.0012673025485128164
Epoch  261 MSE:  0.001263559446670115
Epoch  262 MSE:  0.001259923679754138
Epoch  263 MSE:  0.001256398158147931
Epoch  264 MSE:  0.0012529853265732527
Epoch  265 MSE:  0.0012496879789978266
Epoch  266 MSE:  0.0012465112376958132
Epoch  267 MSE:  0.0012434730306267738
Epoch  268 MSE:  0.0012406742898747325
Epoch  269 MSE:  0.0012388165341690183
Epoch  270 MSE:  0.0012429836206138134
Epoch  271 MSE:  0.0012875766260549426
Epoch  272 MSE:  0.001492623589001596
Epoch  273 MSE:  0.00201231031678617
Epoch  274 MSE:  0.0016974164173007011
Epoch  275 MSE:  0.0012310986639931798
Epoch  276 MSE:  0.001519946032203734
Epoch  277 MSE:  0.0012483973987400532
Epoch  278 MSE:  0.0014362889342010021
Epoch  279 MSE:  0.0012453354429453611
Epoch  280 MSE:  0.0013879925245419145
Epoch  281 MSE:  0.0012415138771757483
Epoch  282 MSE:  0.0013531947042793036
Epoch  283 MSE:  0.0012601300841197371
Epoch  284 MSE:  0.0013007718371227384
Epoch  285 MSE:  0.0012898016721010208
Epoch  286 MSE:  0.001256495830602944
Epoch  287 MSE:  0.0013036596355959773
Epoch  288 MSE:  0.0012414618395268917
Epoch  289 MSE:  0.0012859577545896173
Epoch  290 MSE:  0.0012539169983938336
Epoch  291 MSE:  0.0012540979078039527
Epoch  292 MSE:  0.001267384272068739
Epoch  293 MSE:  0.0012348323361948133
Epoch  294 MSE:  0.0012621035566553473
Epoch  295 MSE:  0.0012356534134596586
Epoch  296 MSE:  0.0012432103976607323
Epoch  297 MSE:  0.0012424476444721222
Epoch  298 MSE:  0.001227591885253787
Epoch  299 MSE:  0.0012415907112881541
Epoch  300 MSE:  0.0012228378327563405
Epoch  301 MSE:  0.0012320202076807618
Epoch  302 MSE:  0.0012245419202372432
Epoch  303 MSE:  0.001221132930368185
Epoch  304 MSE:  0.001225392334163189
Epoch  305 MSE:  0.0012146109947934747
Epoch  306 MSE:  0.0012220368953421712
Epoch  307 MSE:  0.0012130143586546183
Epoch  308 MSE:  0.001216199016198516
Epoch  309 MSE:  0.0012134237913414836
Epoch  310 MSE:  0.0012109134113416076
Epoch  311 MSE:  0.0012131142430007458
Epoch  312 MSE:  0.0012078388826921582
Epoch  313 MSE:  0.0012114611454308033
Epoch  314 MSE:  0.0012066643685102463
Epoch  315 MSE:  0.0012091086246073246
Epoch  316 MSE:  0.0012064219918102026
Epoch  317 MSE:  0.0012068765936419368
Epoch  318 MSE:  0.0012063715839758515
Epoch  319 MSE:  0.001205186010338366
Epoch  320 MSE:  0.0012061332818120718
Epoch  321 MSE:  0.0012041557347401977
Epoch  322 MSE:  0.0012056321138516068
Epoch  323 MSE:  0.0012036605039611459
Epoch  324 MSE:  0.001204974832944572
Epoch  325 MSE:  0.001203469350002706
Epoch  326 MSE:  0.0012043254682794213
Epoch  327 MSE:  0.0012033901875838637
Epoch  328 MSE:  0.0012037695851176977
Epoch  329 MSE:  0.0012033282546326518
Epoch  330 MSE:  0.0012033232487738132
Epoch  331 MSE:  0.0012032492086291313
Epoch  332 MSE:  0.001202977728098631
Epoch  333 MSE:  0.0012031331425532699
Epoch  334 MSE:  0.0012027207994833589
Epoch  335 MSE:  0.0012029799399897456
Epoch  336 MSE:  0.001202528947032988
Epoch  337 MSE:  0.0012028022902086377
Epoch  338 MSE:  0.0012023745803162456
Epoch  339 MSE:  0.0012026189360767603
Epoch  340 MSE:  0.0012022380251437426
Epoch  341 MSE:  0.001202437444590032
Epoch  342 MSE:  0.001202112645842135
Epoch  343 MSE:  0.0012022604933008552
Epoch  344 MSE:  0.0012019947171211243
Epoch  345 MSE:  0.001202090410515666
Epoch  346 MSE:  0.0012018806301057339
Epoch  347 MSE:  0.001201931620016694
Epoch  348 MSE:  0.001201768172904849
Epoch  349 MSE:  0.0012017858680337667
Epoch  350 MSE:  0.0012016582768410444
Epoch  351 MSE:  0.001201652456074953
Epoch  352 MSE:  0.00120155222248286
Epoch  353 MSE:  0.001201529405079782
Epoch  354 MSE:  0.0012014509411528707
Epoch  355 MSE:  0.0012014165986329317
Epoch  356 MSE:  0.0012013546656817198
Epoch  357 MSE:  0.0012013136874884367
Epoch  358 MSE:  0.001201263046823442
Epoch  359 MSE:  0.0012012196239084005
Epoch  360 MSE:  0.0012011764338240027
Epoch  361 MSE:  0.0012011328944936395
Epoch  362 MSE:  0.0012010951759293675
Epoch  363 MSE:  0.0012010523350909352
Epoch  364 MSE:  0.0012010184582322836
Epoch  365 MSE:  0.0012009773636236787
Epoch  366 MSE:  0.0012009465135633945
Epoch  367 MSE:  0.0012009072815999389
Epoch  368 MSE:  0.001200877595692873
Epoch  369 MSE:  0.0012008408084511757
Epoch  370 MSE:  0.001200812985189259
Epoch  371 MSE:  0.0012007781770080328
Epoch  372 MSE:  0.0012007508194074035
Epoch  373 MSE:  0.0012007184559479356
Epoch  374 MSE:  0.0012006916804239154
Epoch  375 MSE:  0.001200660946778953
Epoch  376 MSE:  0.0012006351025775075
Epoch  377 MSE:  0.0012006063479930162
Epoch  378 MSE:  0.0012005807366222143
Epoch  379 MSE:  0.0012005536118522286
Epoch  380 MSE:  0.0012005286989733577
Epoch  381 MSE:  0.0012005026219412684
Epoch  382 MSE:  0.0012004779418930411
Epoch  383 MSE:  0.0012004532618448138
Epoch  384 MSE:  0.0012004293967038393
Epoch  385 MSE:  0.0012004057643935084
Epoch  386 MSE:  0.0012003821320831776
Epoch  387 MSE:  0.001200359663926065
Epoch  388 MSE:  0.0012003362644463778
Epoch  389 MSE:  0.0012003148440271616
Epoch  390 MSE:  0.0012002922594547272
Epoch  391 MSE:  0.0012002714211121202
Epoch  392 MSE:  0.0012002494186162949
Epoch  393 MSE:  0.0012002295115962625
Epoch  394 MSE:  0.0012002078583464026
Epoch  395 MSE:  0.0012001884169876575
Epoch  396 MSE:  0.0012001674622297287
Epoch  397 MSE:  0.0012001482537016273
Epoch  398 MSE:  0.0012001279974356294
Epoch  399 MSE:  0.0012001097202301025
Epoch  400 MSE:  0.0012000901624560356
Epoch  401 MSE:  0.0012000715360045433
Epoch  402 MSE:  0.0012000527931377292
Epoch  403 MSE:  0.0012000349815934896
Epoch  404 MSE:  0.001200016587972641
Epoch  405 MSE:  0.0011999986600130796
Epoch  406 MSE:  0.0011999811977148056
Epoch  407 MSE:  0.0011999636190012097
Epoch  408 MSE:  0.0011999463895335793
Epoch  409 MSE:  0.0011999292764812708
Epoch  410 MSE:  0.001199912279844284
Epoch  411 MSE:  0.0011998956324532628
Epoch  412 MSE:  0.0011998791014775634
Epoch  413 MSE:  0.0011998626869171858
Epoch  414 MSE:  0.0011998465051874518
Epoch  415 MSE:  0.0011998304398730397
Epoch  416 MSE:  0.0011998146073892713
Epoch  417 MSE:  0.001199798658490181
Epoch  418 MSE:  0.0011997830588370562
Epoch  419 MSE:  0.0011997671099379659
Epoch  420 MSE:  0.001199752208776772
Epoch  421 MSE:  0.001199736725538969
Epoch  422 MSE:  0.0011997215915471315
Epoch  423 MSE:  0.0011997065739706159
Epoch  424 MSE:  0.001199691672809422
Epoch  425 MSE:  0.0011996767716482282
Epoch  426 MSE:  0.0011996619869023561
Epoch  427 MSE:  0.0011996474349871278
Epoch  428 MSE:  0.0011996329994872212
Epoch  429 MSE:  0.0011996185639873147
Epoch  430 MSE:  0.0011996041284874082
Epoch  431 MSE:  0.0011995899258181453
Epoch  432 MSE:  0.0011995757231488824
Epoch  433 MSE:  0.0011995616368949413
Epoch  434 MSE:  0.001199547667056322
Epoch  435 MSE:  0.0011995336972177029
Epoch  436 MSE:  0.0011995199602097273
Epoch  437 MSE:  0.0011995062232017517
Epoch  438 MSE:  0.0011994928354397416
Epoch  439 MSE:  0.0011994787491858006
Epoch  440 MSE:  0.0011994653614237905
Epoch  441 MSE:  0.0011994517408311367
Epoch  442 MSE:  0.0011994385858997703
Epoch  443 MSE:  0.0011994248488917947
Epoch  444 MSE:  0.00119941181037575
Epoch  445 MSE:  0.00119939842261374
Epoch  446 MSE:  0.0011993851512670517
Epoch  447 MSE:  0.0011993723455816507
Epoch  448 MSE:  0.0011993593070656061
Epoch  449 MSE:  0.0011993460357189178
Epoch  450 MSE:  0.0011993328807875514
Epoch  451 MSE:  0.001199320307932794
Epoch  452 MSE:  0.0011993072694167495
Epoch  453 MSE:  0.0011992944637313485
Epoch  454 MSE:  0.0011992815416306257
Epoch  455 MSE:  0.0011992689687758684
Epoch  456 MSE:  0.0011992561630904675
Epoch  457 MSE:  0.001199243706651032
Epoch  458 MSE:  0.0011992307845503092
Epoch  459 MSE:  0.0011992183281108737
Epoch  460 MSE:  0.0011992056388407946
Epoch  461 MSE:  0.001199193182401359
Epoch  462 MSE:  0.00119918049313128
Epoch  463 MSE:  0.0011991680366918445
Epoch  464 MSE:  0.0011991556966677308
Epoch  465 MSE:  0.0011991433566436172
Epoch  466 MSE:  0.0011991310166195035
Epoch  467 MSE:  0.001199118560180068
Epoch  468 MSE:  0.0011991062201559544
Epoch  469 MSE:  0.0011990941129624844
Epoch  470 MSE:  0.001199081540107727
Epoch  471 MSE:  0.0011990695493295789
Epoch  472 MSE:  0.0011990572093054652
Epoch  473 MSE:  0.0011990451021119952
Epoch  474 MSE:  0.0011990328785032034
Epoch  475 MSE:  0.0011990207713097334
Epoch  476 MSE:  0.0011990084312856197
Epoch  477 MSE:  0.0011989963240921497
Epoch  478 MSE:  0.0011989843333140016
Epoch  479 MSE:  0.0011989721097052097
Epoch  480 MSE:  0.0011989601189270616
Epoch  481 MSE:  0.0011989480117335916
Epoch  482 MSE:  0.0011989361373707652
Epoch  483 MSE:  0.001198924146592617
Epoch  484 MSE:  0.0011989121558144689
Epoch  485 MSE:  0.0011989000486209989
Epoch  486 MSE:  0.0011988877085968852
Epoch  487 MSE:  0.0011988759506493807
Epoch  488 MSE:  0.0011988637270405889
Epoch  489 MSE:  0.0011988519690930843
Epoch  490 MSE:  0.0011988397454842925
Epoch  491 MSE:  0.0011988281039521098
Epoch  492 MSE:  0.0011988162295892835
Epoch  493 MSE:  0.0011988041223958135
Epoch  494 MSE:  0.0011987922480329871
Epoch  495 MSE:  0.0011987803736701608
Epoch  496 MSE:  0.0011987684993073344
Epoch  497 MSE:  0.0011987563921138644
Epoch  498 MSE:  0.001198744517751038
Epoch  499 MSE:  0.0011987326433882117
Training time: 11.960481882095337

predict = pd.DataFrame(scaler.inverse_transform(y_train_pred.cpu().detach().numpy()))
original = pd.DataFrame(scaler.inverse_transform(y_train_lstm.cpu().detach().numpy()))

sns.set_style("darkgrid")    

fig = plt.figure()
fig.subplots_adjust(hspace=0.2, wspace=0.2)

plt.subplot(1, 2, 1)
ax = sns.lineplot(x = original.index, y = original[0], label="Data", color='royalblue')
ax = sns.lineplot(x = predict.index, y = predict[0], label="Training Prediction (LSTM)", color='tomato')
ax.set_title('Stock price', size = 14, fontweight='bold')
ax.set_xlabel("Days", size = 14)
ax.set_ylabel("Cost (USD)", size = 14)
ax.set_xticklabels('', size=10)


plt.subplot(1, 2, 2)
ax = sns.lineplot(data=hist, color='royalblue')
ax.set_xlabel("Epoch", size = 14)
ax.set_ylabel("Loss", size = 14)
ax.set_title("Training Loss", size = 14, fontweight='bold')
fig.set_figheight(6)
fig.set_figwidth(16)

device = torch.device("cpu")
model.to(device)

x_train = x_train.to(device)
y_train_pred = y_train_pred.to(device)
y_train_lstm = y_train_lstm.to(device)


# make predictions
y_test_pred = model(x_test).cpu()

# invert predictions
y_train_pred = scaler.inverse_transform(y_train_pred.detach().numpy())
y_train = scaler.inverse_transform(y_train_lstm.detach().numpy())
y_test_pred = scaler.inverse_transform(y_test_pred.detach().numpy())
y_test = scaler.inverse_transform(y_test_lstm.detach().numpy())

# calculate root mean squared error
trainScore = math.sqrt(mean_squared_error(y_train[:,0], y_train_pred[:,0]))
print('Train Score: %.2f RMSE' % (trainScore))
testScore = math.sqrt(mean_squared_error(y_test[:,0], y_test_pred[:,0]))
print('Test Score: %.2f RMSE' % (testScore))
lstm.append(trainScore)
lstm.append(testScore)
lstm.append(training_time)

Train Score: 0.51 RMSE
Test Score: 0.93 RMSE

# shift train predictions for plotting
trainPredictPlot = np.empty_like(price)
trainPredictPlot[:, :] = np.nan
trainPredictPlot[lookback:len(y_train_pred)+lookback, :] = y_train_pred

# shift test predictions for plotting
testPredictPlot = np.empty_like(price)
testPredictPlot[:, :] = np.nan
testPredictPlot[len(y_train_pred)+lookback-1:len(price)-1, :] = y_test_pred

original = scaler.inverse_transform(price['Adj Close'].values.reshape(-1,1))

predictions = np.append(trainPredictPlot, testPredictPlot, axis=1)
predictions = np.append(predictions, original, axis=1)
result = pd.DataFrame(predictions)

fig = go.Figure()
fig.add_trace(go.Scatter(go.Scatter(x=df["Date"], y=result[0],
                    mode='lines',
                    name='Train prediction')))
fig.add_trace(go.Scatter(x=df["Date"], y=result[1],
                    mode='lines',
                    name='Test prediction'))
fig.add_trace(go.Scatter(go.Scatter(x=df["Date"], y=result[2],
                    mode='lines',
                    name='Actual Value')))

fig.update_xaxes(
    rangeslider_visible=True,
    rangeselector=dict(
        buttons=list([
            dict(count=1, label="1m", step="month", stepmode="backward"),
            dict(count=6, label="6m", step="month", stepmode="backward"),
            dict(count=1, label="YTD", step="year", stepmode="todate"),
            dict(count=1, label="1y", step="year", stepmode="backward"),
            dict(count=1, label="2y", step="year", stepmode="backward"),
            dict(step="all")
        ]),
        font=dict(family='Rockwell',color='black'),
    )
    )

fig.show()

Build a GRU model

class GRU(nn.Module):
    def __init__(self, input_dim, hidden_dim, num_layers, output_dim):
        super(GRU, self).__init__()
        self.hidden_dim = hidden_dim
        self.num_layers = num_layers
        
        self.gru = nn.GRU(input_dim, hidden_dim, num_layers, batch_first=True)
        self.fc = nn.Linear(hidden_dim, output_dim)

    def forward(self, x):
        h0 = torch.zeros(self.num_layers, x.size(0), self.hidden_dim).requires_grad_().to(device)
        out, (hn) = self.gru(x, (h0.detach()))
        out = self.fc(out[:, -1, :]) 
        return out

model = GRU(input_dim=input_dim, hidden_dim=hidden_dim, output_dim=output_dim, num_layers=num_layers)
criterion = torch.nn.MSELoss(reduction='mean')
optimiser = torch.optim.Adam(model.parameters(), lr=0.01)

model = model.to(device)

criterion.to(device)
x_train = x_train.to(device)
y_train_gru = y_train_gru.to(device)

hist = np.zeros(num_epochs)
start_time = time.time()
gru = []

for t in range(num_epochs):
    y_train_pred = model(x_train)    
    
    loss = criterion(y_train_pred, y_train_gru)
    print("Epoch ", t, "MSE: ", loss.item())
    hist[t] = loss.item()    
    
    optimiser.zero_grad()
    loss.backward()
    optimiser.step()
    
training_time = time.time()-start_time
print("Training time: {}".format(training_time))

Epoch  0 MSE:  0.09333141893148422
Epoch  1 MSE:  0.14284326136112213
Epoch  2 MSE:  0.04148746281862259
Epoch  3 MSE:  0.08071506023406982
Epoch  4 MSE:  0.031027598306536674
Epoch  5 MSE:  0.011289249174296856
Epoch  6 MSE:  0.019751261919736862
Epoch  7 MSE:  0.022235073149204254
Epoch  8 MSE:  0.010761172510683537
Epoch  9 MSE:  0.005481211934238672
Epoch  10 MSE:  0.007945975288748741
Epoch  11 MSE:  0.01044232677668333
Epoch  12 MSE:  0.011669597588479519
Epoch  13 MSE:  0.01093620341271162
Epoch  14 MSE:  0.00786434393376112
Epoch  15 MSE:  0.0041869031265378
Epoch  16 MSE:  0.0030486483592540026
Epoch  17 MSE:  0.005431350786238909
Epoch  18 MSE:  0.007107957731932402
Epoch  19 MSE:  0.005823053885251284
Epoch  20 MSE:  0.004497723188251257
Epoch  21 MSE:  0.0043601300567388535
Epoch  22 MSE:  0.004275926388800144
Epoch  23 MSE:  0.003639059141278267
Epoch  24 MSE:  0.0031692786142230034
Epoch  25 MSE:  0.0037069960962980986
Epoch  26 MSE:  0.004504668992012739
Epoch  27 MSE:  0.004151158966124058
Epoch  28 MSE:  0.0032008399721235037
Epoch  29 MSE:  0.0028251002077013254
Epoch  30 MSE:  0.0029367287643253803
Epoch  31 MSE:  0.0029540611431002617
Epoch  32 MSE:  0.0027779985684901476
Epoch  33 MSE:  0.002738407114520669
Epoch  34 MSE:  0.0029480746015906334
Epoch  35 MSE:  0.0030101609881967306
Epoch  36 MSE:  0.0026732529513537884
Epoch  37 MSE:  0.002299028914421797
Epoch  38 MSE:  0.002244044793769717
Epoch  39 MSE:  0.002394381444901228
Epoch  40 MSE:  0.0024728148709982634
Epoch  41 MSE:  0.002410072134807706
Epoch  42 MSE:  0.002331031020730734
Epoch  43 MSE:  0.002305177040398121
Epoch  44 MSE:  0.0022459591273218393
Epoch  45 MSE:  0.0021093646064400673
Epoch  46 MSE:  0.0020122758578509092
Epoch  47 MSE:  0.002040587831288576
Epoch  48 MSE:  0.002114779083058238
Epoch  49 MSE:  0.002115062205120921
Epoch  50 MSE:  0.0020325342193245888
Epoch  51 MSE:  0.0019503985531628132
Epoch  52 MSE:  0.0019168432336300611
Epoch  53 MSE:  0.0018990904791280627
Epoch  54 MSE:  0.0018699258798733354
Epoch  55 MSE:  0.0018547893268987536
Epoch  56 MSE:  0.0018639949848875403
Epoch  57 MSE:  0.001859020907431841
Epoch  58 MSE:  0.0018133263802155852
Epoch  59 MSE:  0.0017564125591889024
Epoch  60 MSE:  0.0017312460113316774
Epoch  61 MSE:  0.0017338049365207553
Epoch  62 MSE:  0.0017293855780735612
Epoch  63 MSE:  0.001710184384137392
Epoch  64 MSE:  0.0016924866940826178
Epoch  65 MSE:  0.0016767147462815046
Epoch  66 MSE:  0.0016523788217455149
Epoch  67 MSE:  0.0016270113410428166
Epoch  68 MSE:  0.0016172248870134354
Epoch  69 MSE:  0.0016179620288312435
Epoch  70 MSE:  0.0016088931588456035
Epoch  71 MSE:  0.0015877644764259458
Epoch  72 MSE:  0.0015694905305281281
Epoch  73 MSE:  0.001558594754897058
Epoch  74 MSE:  0.0015478333225473762
Epoch  75 MSE:  0.0015364120481535792
Epoch  76 MSE:  0.001528959721326828
Epoch  77 MSE:  0.0015219537308439612
Epoch  78 MSE:  0.0015088655054569244
Epoch  79 MSE:  0.0014938311651349068
Epoch  80 MSE:  0.0014847107231616974
Epoch  81 MSE:  0.0014796123141422868
Epoch  82 MSE:  0.001472295611165464
Epoch  83 MSE:  0.0014628851786255836
Epoch  84 MSE:  0.0014543880242854357
Epoch  85 MSE:  0.0014460552483797073
Epoch  86 MSE:  0.0014371885918080807
Epoch  87 MSE:  0.001430271309800446
Epoch  88 MSE:  0.0014254908310249448
Epoch  89 MSE:  0.0014194028917700052
Epoch  90 MSE:  0.0014112520730122924
Epoch  91 MSE:  0.0014039549278095365
Epoch  92 MSE:  0.0013984640827402472
Epoch  93 MSE:  0.0013930745190009475
Epoch  94 MSE:  0.0013874280266463757
Epoch  95 MSE:  0.0013821744360029697
Epoch  96 MSE:  0.0013766869669780135
Epoch  97 MSE:  0.0013706929748877883
Epoch  98 MSE:  0.0013654601061716676
Epoch  99 MSE:  0.0013612997718155384
Epoch  100 MSE:  0.0013568734284490347
Epoch  101 MSE:  0.0013518015621230006
Epoch  102 MSE:  0.0013470025733113289
Epoch  103 MSE:  0.0013426828663796186
Epoch  104 MSE:  0.0013384444173425436
Epoch  105 MSE:  0.0013343938626348972
Epoch  106 MSE:  0.0013305442407727242
Epoch  107 MSE:  0.0013264798326417804
Epoch  108 MSE:  0.001322334399446845
Epoch  109 MSE:  0.0013186425203457475
Epoch  110 MSE:  0.001315259956754744
Epoch  111 MSE:  0.0013117281487211585
Epoch  112 MSE:  0.0013081388315185905
Epoch  113 MSE:  0.0013047363609075546
Epoch  114 MSE:  0.0013014484429731965
Epoch  115 MSE:  0.001298255636356771
Epoch  116 MSE:  0.0012952324468642473
Epoch  117 MSE:  0.001292238594032824
Epoch  118 MSE:  0.0012891854858025908
Epoch  119 MSE:  0.0012862655567005277
Epoch  120 MSE:  0.0012835602974519134
Epoch  121 MSE:  0.0012809012550860643
Epoch  122 MSE:  0.001278234412893653
Epoch  123 MSE:  0.0012756555806845427
Epoch  124 MSE:  0.0012731729075312614
Epoch  125 MSE:  0.0012707647401839495
Epoch  126 MSE:  0.0012684576213359833
Epoch  127 MSE:  0.0012662160443142056
Epoch  128 MSE:  0.0012639909982681274
Epoch  129 MSE:  0.0012618412729352713
Epoch  130 MSE:  0.0012598125031217933
Epoch  131 MSE:  0.0012578460155054927
Epoch  132 MSE:  0.001255908515304327
Epoch  133 MSE:  0.0012540369061753154
Epoch  134 MSE:  0.0012522421311587095
Epoch  135 MSE:  0.0012505091726779938
Epoch  136 MSE:  0.0012488430365920067
Epoch  137 MSE:  0.001247233827598393
Epoch  138 MSE:  0.0012456653639674187
Epoch  139 MSE:  0.001244157087057829
Epoch  140 MSE:  0.001242722850292921
Epoch  141 MSE:  0.0012413384392857552
Epoch  142 MSE:  0.0012399920960888267
Epoch  143 MSE:  0.0012386994203552604
Epoch  144 MSE:  0.0012374641373753548
Epoch  145 MSE:  0.0012362755369395018
Epoch  146 MSE:  0.0012351330369710922
Epoch  147 MSE:  0.0012340336106717587
Epoch  148 MSE:  0.0012329750461503863
Epoch  149 MSE:  0.0012319623492658138
Epoch  150 MSE:  0.0012309955200180411
Epoch  151 MSE:  0.0012300637317821383
Epoch  152 MSE:  0.001229165936820209
Epoch  153 MSE:  0.0012283088872209191
Epoch  154 MSE:  0.0012274902546778321
Epoch  155 MSE:  0.001226703287102282
Epoch  156 MSE:  0.001225948566570878
Epoch  157 MSE:  0.0012252257438376546
Epoch  158 MSE:  0.0012245336547493935
Epoch  159 MSE:  0.0012238724157214165
Epoch  160 MSE:  0.0012232392327859998
Epoch  161 MSE:  0.0012226310791447759
Epoch  162 MSE:  0.0012220497010275722
Epoch  163 MSE:  0.0012214949820190668
Epoch  164 MSE:  0.0012209640117362142
Epoch  165 MSE:  0.0012204546947032213
Epoch  166 MSE:  0.001219967845827341
Epoch  167 MSE:  0.001219503115862608
Epoch  168 MSE:  0.001219058409333229
Epoch  169 MSE:  0.001218633260577917
Epoch  170 MSE:  0.0012182266218587756
Epoch  171 MSE:  0.0012178380275145173
Epoch  172 MSE:  0.0012174672447144985
Epoch  173 MSE:  0.0012171129928901792
Epoch  174 MSE:  0.0012167737586423755
Epoch  175 MSE:  0.0012164500076323748
Epoch  176 MSE:  0.001216141157783568
Epoch  177 MSE:  0.0012158462777733803
Epoch  178 MSE:  0.0012155642034485936
Epoch  179 MSE:  0.0012152951676398516
Epoch  180 MSE:  0.0012150387046858668
Epoch  181 MSE:  0.0012147933011874557
Epoch  182 MSE:  0.0012145593063905835
Epoch  183 MSE:  0.0012143359053879976
Epoch  184 MSE:  0.0012141227489337325
Epoch  185 MSE:  0.001213919254951179
Epoch  186 MSE:  0.0012137250741943717
Epoch  187 MSE:  0.0012135393917560577
Epoch  188 MSE:  0.0012133622076362371
Epoch  189 MSE:  0.0012131932890042663
Epoch  190 MSE:  0.0012130315881222486
Epoch  191 MSE:  0.0012128771049901843
Epoch  192 MSE:  0.0012127296067774296
Epoch  193 MSE:  0.0012125886278226972
Epoch  194 MSE:  0.001212453586049378
Epoch  195 MSE:  0.0012123244814574718
Epoch  196 MSE:  0.0012122004991397262
Epoch  197 MSE:  0.00121208222117275
Epoch  198 MSE:  0.0012119687162339687
Epoch  199 MSE:  0.0012118594022467732
Epoch  200 MSE:  0.0012117548612877727
Epoch  201 MSE:  0.0012116542784497142
Epoch  202 MSE:  0.0012115574209019542
Epoch  203 MSE:  0.0012114644050598145
Epoch  204 MSE:  0.001211374532431364
Epoch  205 MSE:  0.0012112879194319248
Epoch  206 MSE:  0.001211204333230853
Epoch  207 MSE:  0.0012111234245821834
Epoch  208 MSE:  0.0012110451934859157
Epoch  209 MSE:  0.0012109694071114063
Epoch  210 MSE:  0.0012108958326280117
Epoch  211 MSE:  0.0012108244700357318
Epoch  212 MSE:  0.0012107548536732793
Epoch  213 MSE:  0.0012106872163712978
Epoch  214 MSE:  0.001210621208883822
Epoch  215 MSE:  0.0012105568312108517
Epoch  216 MSE:  0.0012104939669370651
Epoch  217 MSE:  0.0012104324996471405
Epoch  218 MSE:  0.0012103721965104342
Epoch  219 MSE:  0.001210313057526946
Epoch  220 MSE:  0.0012102550826966763
Epoch  221 MSE:  0.0012101982720196247
Epoch  222 MSE:  0.001210142276249826
Epoch  223 MSE:  0.0012100872118026018
Epoch  224 MSE:  0.0012100327294319868
Epoch  225 MSE:  0.0012099792947992682
Epoch  226 MSE:  0.0012099264422431588
Epoch  227 MSE:  0.0012098741717636585
Epoch  228 MSE:  0.0012098224833607674
Epoch  229 MSE:  0.0012097714934498072
Epoch  230 MSE:  0.0012097208527848125
Epoch  231 MSE:  0.0012096707941964269
Epoch  232 MSE:  0.001209620968438685
Epoch  233 MSE:  0.0012095717247575521
Epoch  234 MSE:  0.0012095228303223848
Epoch  235 MSE:  0.0012094741687178612
Epoch  236 MSE:  0.001209425856359303
Epoch  237 MSE:  0.0012093778932467103
Epoch  238 MSE:  0.0012093300465494394
Epoch  239 MSE:  0.0012092826655134559
Epoch  240 MSE:  0.0012092354008927941
Epoch  241 MSE:  0.001209188369102776
Epoch  242 MSE:  0.001209141337312758
Epoch  243 MSE:  0.0012090946547687054
Epoch  244 MSE:  0.0012090482050552964
Epoch  245 MSE:  0.0012090018717572093
Epoch  246 MSE:  0.0012089555384591222
Epoch  247 MSE:  0.0012089093215763569
Epoch  248 MSE:  0.0012088633375242352
Epoch  249 MSE:  0.0012088174698874354
Epoch  250 MSE:  0.0012087714858353138
Epoch  251 MSE:  0.001208725618198514
Epoch  252 MSE:  0.0012086800998076797
Epoch  253 MSE:  0.0012086343485862017
Epoch  254 MSE:  0.0012085885973647237
Epoch  255 MSE:  0.001208543311804533
Epoch  256 MSE:  0.0012084976769983768
Epoch  257 MSE:  0.0012084522750228643
Epoch  258 MSE:  0.00120840675663203
Epoch  259 MSE:  0.0012083611218258739
Epoch  260 MSE:  0.0012083158362656832
Epoch  261 MSE:  0.0012082704342901707
Epoch  262 MSE:  0.00120822514872998
Epoch  263 MSE:  0.0012081797467544675
Epoch  264 MSE:  0.0012081342283636332
Epoch  265 MSE:  0.0012080888263881207
Epoch  266 MSE:  0.0012080433079972863
Epoch  267 MSE:  0.0012079979060217738
Epoch  268 MSE:  0.0012079525040462613
Epoch  269 MSE:  0.0012079068692401052
Epoch  270 MSE:  0.001207861234433949
Epoch  271 MSE:  0.0012078157160431147
Epoch  272 MSE:  0.0012077700812369585
Epoch  273 MSE:  0.0012077243300154805
Epoch  274 MSE:  0.0012076786952093244
Epoch  275 MSE:  0.0012076328275725245
Epoch  276 MSE:  0.0012075870763510466
Epoch  277 MSE:  0.001207541092298925
Epoch  278 MSE:  0.0012074949918314815
Epoch  279 MSE:  0.0012074491241946816
Epoch  280 MSE:  0.0012074029073119164
Epoch  281 MSE:  0.001207356690429151
Epoch  282 MSE:  0.001207310357131064
Epoch  283 MSE:  0.0012072641402482986
Epoch  284 MSE:  0.0012072176905348897
Epoch  285 MSE:  0.0012071712408214808
Epoch  286 MSE:  0.0012071247911080718
Epoch  287 MSE:  0.0012070779921486974
Epoch  288 MSE:  0.0012070313096046448
Epoch  289 MSE:  0.0012069842778146267
Epoch  290 MSE:  0.0012069374788552523
Epoch  291 MSE:  0.0012068904470652342
Epoch  292 MSE:  0.001206843531690538
Epoch  293 MSE:  0.0012067961506545544
Epoch  294 MSE:  0.0012067487696185708
Epoch  295 MSE:  0.0012067012721672654
Epoch  296 MSE:  0.0012066536583006382
Epoch  297 MSE:  0.001206606044434011
Epoch  298 MSE:  0.0012065580813214183
Epoch  299 MSE:  0.0012065102346241474
Epoch  300 MSE:  0.0012064622715115547
Epoch  301 MSE:  0.0012064140755683184
Epoch  302 MSE:  0.0012063659960404038
Epoch  303 MSE:  0.0012063175672665238
Epoch  304 MSE:  0.001206269022077322
Epoch  305 MSE:  0.0012062203604727983
Epoch  306 MSE:  0.0012061716988682747
Epoch  307 MSE:  0.0012061228044331074
Epoch  308 MSE:  0.0012060737935826182
Epoch  309 MSE:  0.0012060245499014854
Epoch  310 MSE:  0.0012059754226356745
Epoch  311 MSE:  0.001205925946123898
Epoch  312 MSE:  0.0012058764696121216
Epoch  313 MSE:  0.0012058268766850233
Epoch  314 MSE:  0.0012057770509272814
Epoch  315 MSE:  0.0012057271087542176
Epoch  316 MSE:  0.001205677050165832
Epoch  317 MSE:  0.0012056267587468028
Epoch  318 MSE:  0.0012055765837430954
Epoch  319 MSE:  0.0012055259430781007
Epoch  320 MSE:  0.0012054751859977841
Epoch  321 MSE:  0.0012054244289174676
Epoch  322 MSE:  0.0012053734390065074
Epoch  323 MSE:  0.0012053223326802254
Epoch  324 MSE:  0.0012052709935232997
Epoch  325 MSE:  0.0012052195379510522
Epoch  326 MSE:  0.0012051679659634829
Epoch  327 MSE:  0.0012051162775605917
Epoch  328 MSE:  0.0012050643563270569
Epoch  329 MSE:  0.0012050123186782002
Epoch  330 MSE:  0.0012049601646140218
Epoch  331 MSE:  0.0012049076613038778
Epoch  332 MSE:  0.0012048552744090557
Epoch  333 MSE:  0.001204802538268268
Epoch  334 MSE:  0.0012047495692968369
Epoch  335 MSE:  0.0012046964839100838
Epoch  336 MSE:  0.0012046433985233307
Epoch  337 MSE:  0.0012045899638906121
Epoch  338 MSE:  0.0012045365292578936
Epoch  339 MSE:  0.0012044827453792095
Epoch  340 MSE:  0.0012044287286698818
Epoch  341 MSE:  0.001204374828375876
Epoch  342 MSE:  0.0012043205788359046
Epoch  343 MSE:  0.0012042660964652896
Epoch  344 MSE:  0.0012042114976793528
Epoch  345 MSE:  0.001204156898893416
Epoch  346 MSE:  0.0012041018344461918
Epoch  347 MSE:  0.0012040467699989676
Epoch  348 MSE:  0.001203991356305778
Epoch  349 MSE:  0.0012039359426125884
Epoch  350 MSE:  0.001203880412504077
Epoch  351 MSE:  0.0012038244167342782
Epoch  352 MSE:  0.0012037684209644794
Epoch  353 MSE:  0.0012037124251946807
Epoch  354 MSE:  0.0012036558473482728
Epoch  355 MSE:  0.0012035993859171867
Epoch  356 MSE:  0.0012035425752401352
Epoch  357 MSE:  0.0012034856481477618
Epoch  358 MSE:  0.0012034286046400666
Epoch  359 MSE:  0.001203371211886406
Epoch  360 MSE:  0.0012033137027174234
Epoch  361 MSE:  0.001203256193548441
Epoch  362 MSE:  0.001203198335133493
Epoch  363 MSE:  0.0012031401274725795
Epoch  364 MSE:  0.0012030820362269878
Epoch  365 MSE:  0.0012030235957354307
Epoch  366 MSE:  0.0012029648059979081
Epoch  367 MSE:  0.0012029061326757073
Epoch  368 MSE:  0.0012028469936922193
Epoch  369 MSE:  0.0012027877382934093
Epoch  370 MSE:  0.0012027284828945994
Epoch  371 MSE:  0.0012026689946651459
Epoch  372 MSE:  0.0012026091571897268
Epoch  373 MSE:  0.0012025493197143078
Epoch  374 MSE:  0.0012024890165776014
Epoch  375 MSE:  0.001202428713440895
Epoch  376 MSE:  0.0012023680610582232
Epoch  377 MSE:  0.0012023074086755514
Epoch  378 MSE:  0.001202246407046914
Epoch  379 MSE:  0.001202185289002955
Epoch  380 MSE:  0.0012021239381283522
Epoch  381 MSE:  0.0012020623544231057
Epoch  382 MSE:  0.0012020006543025374
Epoch  383 MSE:  0.0012019387213513255
Epoch  384 MSE:  0.00120187655556947
Epoch  385 MSE:  0.0012018142733722925
Epoch  386 MSE:  0.0012017517583444715
Epoch  387 MSE:  0.0012016890104860067
Epoch  388 MSE:  0.0012016261462122202
Epoch  389 MSE:  0.0012015631655231118
Epoch  390 MSE:  0.0012014999520033598
Epoch  391 MSE:  0.0012014363892376423
Epoch  392 MSE:  0.001201372710056603
Epoch  393 MSE:  0.0012013086816295981
Epoch  394 MSE:  0.0012012446532025933
Epoch  395 MSE:  0.001201180275529623
Epoch  396 MSE:  0.0012011158978566527
Epoch  397 MSE:  0.0012010512873530388
Epoch  398 MSE:  0.0012009863276034594
Epoch  399 MSE:  0.00120092136785388
Epoch  400 MSE:  0.001200856058858335
Epoch  401 MSE:  0.00120079074986279
Epoch  402 MSE:  0.0012007249752059579
Epoch  403 MSE:  0.0012006593169644475
Epoch  404 MSE:  0.0012005933094769716
Epoch  405 MSE:  0.001200527185574174
Epoch  406 MSE:  0.0012004607124254107
Epoch  407 MSE:  0.0012003942392766476
Epoch  408 MSE:  0.001200327416881919
Epoch  409 MSE:  0.0012002605944871902
Epoch  410 MSE:  0.001200193422846496
Epoch  411 MSE:  0.001200126251205802
Epoch  412 MSE:  0.0012000587303191423
Epoch  413 MSE:  0.001199991093017161
Epoch  414 MSE:  0.0011999233392998576
Epoch  415 MSE:  0.0011998553527519107
Epoch  416 MSE:  0.0011997870169579983
Epoch  417 MSE:  0.0011997186811640859
Epoch  418 MSE:  0.0011996503453701735
Epoch  419 MSE:  0.0011995816603302956
Epoch  420 MSE:  0.001199512742459774
Epoch  421 MSE:  0.0011994438245892525
Epoch  422 MSE:  0.0011993745574727654
Epoch  423 MSE:  0.0011993051739409566
Epoch  424 MSE:  0.001199235673993826
Epoch  425 MSE:  0.0011991660576313734
Epoch  426 MSE:  0.001199096324853599
Epoch  427 MSE:  0.001199026475660503
Epoch  428 MSE:  0.0011989561608061194
Epoch  429 MSE:  0.0011988860787823796
Epoch  430 MSE:  0.0011988157639279962
Epoch  431 MSE:  0.0011987450998276472
Epoch  432 MSE:  0.0011986744357272983
Epoch  433 MSE:  0.0011986036552116275
Epoch  434 MSE:  0.0011985327582806349
Epoch  435 MSE:  0.0011984616285189986
Epoch  436 MSE:  0.0011983903823420405
Epoch  437 MSE:  0.0011983191361650825
Epoch  438 MSE:  0.0011982476571574807
Epoch  439 MSE:  0.001198176178149879
Epoch  440 MSE:  0.0011981045827269554
Epoch  441 MSE:  0.0011980326380580664
Epoch  442 MSE:  0.0011979610426351428
Epoch  443 MSE:  0.0011978887487202883
Epoch  444 MSE:  0.0011978168040513992
Epoch  445 MSE:  0.0011977446265518665
Epoch  446 MSE:  0.001197672332637012
Epoch  447 MSE:  0.0011975999223068357
Epoch  448 MSE:  0.001197527744807303
Epoch  449 MSE:  0.0011974552180618048
Epoch  450 MSE:  0.0011973825749009848
Epoch  451 MSE:  0.001197309698909521
Epoch  452 MSE:  0.001197237055748701
Epoch  453 MSE:  0.001197164412587881
Epoch  454 MSE:  0.0011970914201810956
Epoch  455 MSE:  0.0011970184277743101
Epoch  456 MSE:  0.00119694578461349
Epoch  457 MSE:  0.0011968727922067046
Epoch  458 MSE:  0.0011967996833845973
Epoch  459 MSE:  0.0011967268073931336
Epoch  460 MSE:  0.0011966536985710263
Epoch  461 MSE:  0.0011965807061642408
Epoch  462 MSE:  0.0011965074809268117
Epoch  463 MSE:  0.001196434604935348
Epoch  464 MSE:  0.001196361379697919
Epoch  465 MSE:  0.0011962883872911334
Epoch  466 MSE:  0.0011962151620537043
Epoch  467 MSE:  0.0011961422860622406
Epoch  468 MSE:  0.0011960692936554551
Epoch  469 MSE:  0.0011959961848333478
Epoch  470 MSE:  0.0011959233088418841
Epoch  471 MSE:  0.0011958503164350986
Epoch  472 MSE:  0.001195777440443635
Epoch  473 MSE:  0.0011957046808674932
Epoch  474 MSE:  0.0011956319212913513
Epoch  475 MSE:  0.0011955592781305313
Epoch  476 MSE:  0.0011954867513850331
Epoch  477 MSE:  0.001195414224639535
Epoch  478 MSE:  0.0011953418143093586
Epoch  479 MSE:  0.0011952694039791822
Epoch  480 MSE:  0.0011951973428949714
Epoch  481 MSE:  0.0011951252818107605
Epoch  482 MSE:  0.0011950533371418715
Epoch  483 MSE:  0.0011949815088883042
Epoch  484 MSE:  0.0011949097970500588
Epoch  485 MSE:  0.001194838434457779
Epoch  486 MSE:  0.0011947669554501772
Epoch  487 MSE:  0.0011946957092732191
Epoch  488 MSE:  0.0011946248123422265
Epoch  489 MSE:  0.001194553915411234
Epoch  490 MSE:  0.001194483251310885
Epoch  491 MSE:  0.0011944128200411797
Epoch  492 MSE:  0.0011943423887714744
Epoch  493 MSE:  0.0011942726559937
Epoch  494 MSE:  0.0011942025739699602
Epoch  495 MSE:  0.0011941331904381514
Epoch  496 MSE:  0.0011940636904910207
Epoch  497 MSE:  0.0011939946562051773
Epoch  498 MSE:  0.0011939258547499776
Epoch  499 MSE:  0.0011938571697100997
Training time: 19.867424726486206

predict = pd.DataFrame(scaler.inverse_transform(y_train_pred.cpu().detach().numpy()))
original = pd.DataFrame(scaler.inverse_transform(y_train_gru.cpu().detach().numpy()))

sns.set_style("darkgrid")    

fig = plt.figure()
fig.subplots_adjust(hspace=0.2, wspace=0.2)

plt.subplot(1, 2, 1)
ax = sns.lineplot(x = original.index, y = original[0], label="Data", color='royalblue')
ax = sns.lineplot(x = predict.index, y = predict[0], label="Training Prediction (GRU)", color='tomato')
ax.set_title('Stock price', size = 14, fontweight='bold')
ax.set_xlabel("Days", size = 14)
ax.set_ylabel("Cost (USD)", size = 14)
ax.set_xticklabels('', size=10)


plt.subplot(1, 2, 2)
ax = sns.lineplot(data=hist, color='royalblue')
ax.set_xlabel("Epoch", size = 14)
ax.set_ylabel("Loss", size = 14)
ax.set_title("Training Loss", size = 14, fontweight='bold')
fig.set_figheight(6)
fig.set_figwidth(16)

device = torch.device("cpu")
model.to(device)

x_train = x_train.to(device)
y_train_pred = y_train_pred.to(device)
y_train_gru = y_train_gru.to(device)


# make predictions
y_test_pred = model(x_test).cpu()

# invert predictions
y_train_pred = scaler.inverse_transform(y_train_pred.detach().numpy())
y_train = scaler.inverse_transform(y_train_gru.detach().numpy())
y_test_pred = scaler.inverse_transform(y_test_pred.detach().numpy())
y_test = scaler.inverse_transform(y_test_gru.detach().numpy())

# calculate root mean squared error
trainScore = math.sqrt(mean_squared_error(y_train[:,0], y_train_pred[:,0]))
print('Train Score: %.2f RMSE' % (trainScore))
testScore = math.sqrt(mean_squared_error(y_test[:,0], y_test_pred[:,0]))
print('Test Score: %.2f RMSE' % (testScore))
gru.append(trainScore)
gru.append(testScore)
gru.append(training_time)

Train Score: 0.51 RMSE
Test Score: 0.94 RMSE

# shift train predictions for plotting
trainPredictPlot = np.empty_like(price)
trainPredictPlot[:, :] = np.nan
trainPredictPlot[lookback:len(y_train_pred)+lookback, :] = y_train_pred

# shift test predictions for plotting
testPredictPlot = np.empty_like(price)
testPredictPlot[:, :] = np.nan
testPredictPlot[len(y_train_pred)+lookback-1:len(price)-1, :] = y_test_pred

original = scaler.inverse_transform(price['Adj Close'].values.reshape(-1,1))

predictions = np.append(trainPredictPlot, testPredictPlot, axis=1)
predictions = np.append(predictions, original, axis=1)
result = pd.DataFrame(predictions)

fig = go.Figure()
fig.add_trace(go.Scatter(go.Scatter(x=df["Date"], y=result[0],
                    mode='lines',
                    name='Train prediction')))
fig.add_trace(go.Scatter(x=df["Date"], y=result[1],
                    mode='lines',
                    name='Test prediction'))
fig.add_trace(go.Scatter(go.Scatter(x=df["Date"], y=result[2],
                    mode='lines',
                    name='Actual Value')))

fig.update_xaxes(
    rangeslider_visible=True,
    rangeselector=dict(
        buttons=list([
            dict(count=1, label="1m", step="month", stepmode="backward"),
            dict(count=6, label="6m", step="month", stepmode="backward"),
            dict(count=1, label="YTD", step="year", stepmode="todate"),
            dict(count=1, label="1y", step="year", stepmode="backward"),
            dict(count=1, label="2y", step="year", stepmode="backward"),
            dict(step="all")
        ]),
        font=dict(family='Rockwell',color='black'),
    )
    )

fig.show()

References

• https://medium.com/swlh/stock-price-prediction-with-pytorch-37f52ae84632

• https://github.com/udacity/deep-learning-v2-pytorch/blob/master/recurrent-neural-networks/time-series/Simple_RNN.ipynb

• https://medium.com/analytics-vidhya/rnn-vs-gru-vs-lstm-863b0b7b1573